Workshop 3_Timeseries Models

Objectives

The primary objectives of this analysis is to use ARIMA models to find if Photosynthetic Active Radiation is a more suitable explanatory variable to Net Ecosystem Exchange than salinity to forecast future Net Ecosystem Exchange of a scrub mangrove forest.

Methods

Site Information

TS/PH 7 is located 25.19080491, -80.63910514 up Taylor River mainly characterized by scrub mangrove vegetation of the -Rhizophora mangle- species. This is an established FCE LTER site within the Taylor Slough Watershed.

Statistical Analysis

Photosynthetic Active Radiation (par; W m−2 day−1) will be incorporated to the forecast function of the ARIMA model after the temporal dynamics of NEE were adquired. The Augmented Dickey-Fuller Test will be used to test stationarity. Ljung-Box test will test for independece and finally the Akaike Information Criteria (AIC) of the model outputs will be compared to test for model fitness.

Results

ARIMA.NEE4 SALINITY MODEL

Table 1

Model	df	AIC
arima.nee2	18	704.7663
arima.nee5	10	683.6937
arima.nee4	9	700.7734

Discussion

Photosynthetic Active Radiation seems to be a better explanatory series of nee than salinity with a lower AIC as you can see in table 1.